// vfnmsac: vd[i] = -(f[rs1] * vs2[i]) + vd[i]
VI_VFP_VF_LOOP
({
  vd = f32_mulAdd(rs1, f32(vs2.v ^ F32_SIGN), vd);
})